Introduction

Glutathione (γ-glutamyl-l-cysteinylglycine, reduced form GSH) is the major non-protein thiol compound present in living organisms and is involved, predominantly, in buffering redox status of the cytoplasm and protecting cells against oxidative damage (Carmel-Harel and Storz 2000). It is synthesised in two ATP-dependent steps (Meister and Anderson 1983). The dipeptide γ-glutamylcysteine (γ-GC) is first synthesised from l-glutamic acid and l-cysteine by γ-glutamylcysteine synthetase (also known as glutamate-cysteine ligase, EC 6.3.2.2, γ-GCS). In the second step, catalysed by glutathione synthetase (EC 6.3.2.3, GS), glycine is added to the C-terminal end of γ-GC to form GSH. In microorganisms, GSH is found primarily in eukaryotes and in Gram-negative bacteria, but rarely in Gram-positive bacteria (Fahey et al. 1978). Of the latter, GSH is found in some low-GC content Gram-positive bacteria (Streptococcus, Lactococcus, Lactobacillus, Enterococcus, Clostridium and Listeria) (Fahey et al. 1978; Fernández and Steele 1993; Newton et al. 1996). Other Gram-positive bacteria (i.e. Streptococcus thermophilus, Streptococcus agalactiae and Enterococcus faecalis) were reported to synthesise GSH (Newton et al. 1996). However, there is, to the best of our knowledge, no direct evidence showing the presence of the activities of γ-GCS and GS in those strains. In addition, analysis of the genome databases reveals that the genes gshA and gshB, encoding γ-GCS and GS, respectively, are present neither in the genome sequences of S. agalactiae and E. faecalis (KEGG genes database 2004) nor in that of S. thermophilus (Genome sequencing project of S. thermophilus2004), which casts doubts on the conclusions drawn by Newton et al. (1996).

A survey of protein databases like PFAM and NCBI reveals that genes with high similarity to gshA are found in Gram-positive bacteria. However, genes with similarity to gshB have until now not been observed in Gram-positives (see also Copley and Dhillon 2002). The assignment of a function to the corresponding γ-GCS-like hypothetical proteins is uncertain, particularly because these organisms are not known to synthesise GSH (Newton et al. 1993, 1996). These observations do suggest that Gram-positive bacteria generally lack the capability to synthesise glutathione, although some strains are able to import glutathione from the environment (Sherrill and Fahey 1998).

In a previous study (Li et al. 2003), we provided evidence that dairy strains of Lactococcus lactis cannot synthesise glutathione, but are able to take it up from the growth medium. Using L. lactis ssp. cremoris SK11 as a model strain, we found that glutathione taken up by strain SK11 increases its resistance to oxidative stress. As a consequence of these studies we were interested in producing glutathione in L. lactis. This could putatively render the cells more resistant to oxidative stress, but could also have beneficial results on fermentation products made using this organism. For example, it has been shown that flavour development in cheese curds is enhanced when glutathione is added (Singh and Kristoffersen 1971). In the present study, using the nisin-controlled expression system (NICE) for L. lactis (de Ruyter et al. 1996; Kleerebezem et al. 2000), a very high intracellular content of GSH, 358 nmol mg−1 protein, was attained by simultaneously expressing gshA and gshB genes from Escherichia coli. To our knowledge, this is the first example of an engineered Gram-positive bacterium producing glutathione. Furthermore, the glutathione content found in recombinant L. lactis is the highest reported in bacteria. This strain also offers the opportunity to study the effect of in situ glutathione production on the properties of fermented foods.

Materials and methods

Chemicals

GSH, γ-GC, ATP, and all amino acids were purchased from Sigma (St. Louis, Mo.). Monobromobimane (mBBr) was purchased from Calbiochem (San Diego, Calif.). All inorganic compounds were of reagent grade or high quality.

Bacterial strains, plasmids, and media

The strains and plasmids used in this study are listed in Table 1. Escherichia coli TG1, which was used as a chromosome DNA template for amplifying genes gshA and gshB, was grown with aeration in tryptone yeast (TY) extract broth at 37°C for 16 h. Strains of L. lactis were grown without aeration at 30°C in a chemically defined medium (CDM) as described previously (Li et al. 2003) or in M17 broth (Merck, Darmstadt, Germany), supplemented with 5 g l−1 glucose. When appropriate, the medium contained chloramphenicol (10 μg ml−1) as a selection marker. To analyse the effect of gene over-expression, the NICE system was used (de Ruyter et al. 1996; Kleerebezem et al. 2000). For glutathione production and enzyme activity analysis, L. lactis cells were grown until an optical density (OD) at 600 nm of approximately 0.4 was achieved. Subsequently, the culture was split into two cultures. Nisin (2 ng ml−1) was added to one of the two cultures, and both cultures were grown for an additional 4–8 h to reach the stationary phase.

Table 1 Strains and plasmids used in this study. Cmr Chloramphenicol resistant
Full size table

DNA manipulations

Isolation of chromosome DNA of E. coli and standard recombinant DNA techniques were performed as described by Sambrook et al. (1989). Isolation and transformation of L. lactis DNA were performed as previously described (de Vos et al. 1989).

Construction of strains and plasmids

The E. coli gshA gene was amplified by PCR using Pwo DNA polymerase and chromosomal DNA of E. coli TG1 as template DNA with the primers 5′-GAGGCCATGGCAATCCCGGACGTATCACAGGCGC-3′ and 5′-TTCTTCTAGATCAGGCGTGTTTTTCCAGCCACAC-3′. The 1,574-bp PCR product generated was cloned into pNZ8148 using the NcoI and XbaI sites that were introduced by the primers used, yielding the gshA over-expression plasmid pNZ3201.

The E. coli gshB gene was amplified by PCR using Pwo DNA polymerase and chromosomal DNA of E. coli TG1 as template DNA with the primers 5′-TTGGTCTAGAGGAGAAGAATAATGATCAAGCTCG-3′ and 5′-AAGGAAGCTTTTACTGCTGCTGTAAACGTGCTTC-3′. The 982-bp PCR product generated, containing the Shine-Dalgarno sequence upstream of the gshB gene of E. coli, was cloned into pNZ8148 and pNZ3201, respectively, using the XbaI and HindIII sites that were introduced by the primers used, yielding the gshB overexpression plasmid pNZ3202, and the gshA and gshB overexpression plasmid pNZ3203.

Plasmids pNZ3201, pNZ3202 or pNZ3203 were introduced into L. lactis strain NZ9000 by electroporation. Subsequently, thiol producing capacity was evaluated by HPLC.

Preparation of cell-free extracts and protein analysis

Pre-chilled, overnight grown culture (50 ml) was harvested by centrifugation (10,000 g, 10 min, 4°C). Harvested cells were washed twice with ice-cold saline (0.85% NaCl, w/v) and re-suspended in 1 ml 200 mM phosphate buffer (pH 7.0) containing 2 mM EDTA. From this cell suspension, 1 ml was added to a vial with 1 g glass beads and broken using a mini bead beater (FastPrep FP120, Savant, Farmingdale, N.Y.) at 4°C for 30 s, following which the cell debris was removed by centrifugation (10,000 g, 10 min, 4°C), resulting in cell-free extract (CFE). Protein concentrations were determined using a BCA protein assay kit using bovine serum albumin as the standard (Pierce, Rockford, Ill.). For protein analysis lactococcal CFE was mixed with an equal amount of 2-fold concentrated Laemmli buffer and, after heating at 95°C for 10 min, 10 μl of each sample was analysed by sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE).

Thiol assays

mBBr fluorescent labelling and HPLC methods were used to determine the intracellular thiol (cysteine, γ-GC and glutathione) contents according to methods in the literature (Fahey and Newton 1987). A 90 μl aliquot of a cell suspension was added to 200 μl 50 mM HEPPSO (pH 7.5) containing 50% aqueous acetonitrile, 1 mM dithiothreitol, and 5 mM DTPA, following which 5 μl mBBr (50 mM in acetonitrile) was added. The mixture was heated for 30 min at 60°C, and was subsequently acidified by adding 5 μl 1.2 M methanesulfonic acid. The mixture was centrifuged at 20,000 g for 5 min prior to injection to a PLRP-S 300 Å column (4.6 mm ×250 mm) packed with 5 μm reversed-phase material (Polymer Laboratories, Church Stretton, UK), using a Waters 600E HPLC (Waters, Milford, Mass.) equipped with a Dilutor 401 injector (Gilson, Villiers-le-Bel, France) and an FL2000 fluorometer-detector (Spectra Physics, Darmstadt, Germany) (excitation at 390 nm, emission measured at 480 nm). The elution solvent A is 5% (v/v) aqueous acetonitrile, 0.25% (v/v) glacial acetic acid, pH 3.5, and solvent B is 90% acetonitrile and 0.25% glacial acetic acid in water. The following elution profile was used. Isocratic conditions (100% solvent A) were maintained for 2 min, stepping directly to 35% solvent B in 35 min and to 100% solvent B in further 3 min and maintained for 5 min, followed by an immediate return to initial conditions for re-equilibration (15 min) before loading the next sample. Flow rate was maintained at 1 ml min−1 throughout.

Enzyme assays

The activity of γ-GCS was measured by the method of Jackson (1969) with a slight modification. The reaction mixture contained 30 mmol l-glutamate, 30 mmol l-cysteine, 30 mmol ATP, 20 mmol MgSO4·7H2O, 200 mmol KCl, 200 mmol diethanolamine-HCl buffer (pH 9.15), and CFE in a final volume of 500 μl. Incubation was carried out at 37°C for 15 min. The reactions were terminated by adding 825 μl ice-cold 3.2% sulphosalicylic acid. After removing the protein precipitates by centrifugation, 90 μl sample (or control) supernatant was used to measure the γ-GC concentrations using the mBBr fluorescent labelling HPLC method described above.

The activity of GS was measured according to the method of Gushima et al. (1983a). The reaction was carried out in 200 μl of a mixture containing 5.0 mM γ-GC, 10 mM glycine, 10 mM ATP, 10 mM MgCl2, 100 mM Tris-HCl buffer (pH 7.5), and CFE at 37°C for 15 min. The reactions were terminated by adding 500 μl ice-cold 3.2% sulphosalicylic acid. After removing the protein precipitates by centrifugation, the thiol concentrations were measured in 90 μl sample (or control) by the mBBr fluorescent labelling HPLC method.

In the above two assays, the composition of the control was the same as the reaction mixture except that sulphosalicylic acid was added prior to addition of CFE. One unit of specific enzyme activity was defined as the amount of enzyme that catalysed the synthesis of 1 nmol γ-GC or GSH per minute per milligram of protein.

H2O2 treatment

L. lactis NZ9000 cells harbouring plasmids pNZ8148, pNZ3201, pNZ3202, and pNZ3203 were grown anaerobically in CDM at 30°C until an OD600 of approximately 0.4 was achieved, to which 2 ng ml−1 nisin was added to induce gene expression. The induction phase lasted until cell growth reached stationary phase (OD 2.5, usually after 4 h induction). Samples (10 ml) of the cultures were harvested, washed twice in cold saline, and re-suspended in a volume of saline to achieve a final OD600 of 2.5. Of this cell suspension, 1 ml was treated with 10 mM H2O2 for 30 min, pelleted by centrifugation at 10,000 g for 1 min, then washed again with saline to remove the residual H2O2. Cells were re-suspended in the same volume of saline, and either spread on M17 plates at different dilutions to measure survival, or inoculated into CDM with an inoculum size of 1% (v/v) for growth experiments. Colony counting results were the mean of three plates incubated at 30°C for 48 h with standard deviation shown as error bars.

Growth experiments

A 10 μl aliquot of the H2O2-treated and saline-washed cell suspension was inoculated into 1 ml fresh CDM. Subsequently, 200 μl culture was transferred into a well of a microtitre plate (Corning, Corning, N.Y.). Growth was monitored at 600 nm in a kinetic microtitre plate reader (SPECTRAmax PLUS 384; Molecular Devices, Sunnyvale, Calif.). Each growth experiment was carried out in triplicate. The effect of the duration of the lag phase was estimated by measuring the time needed to reach five times the initial optical density. Results represent the mean of these time intervals estimated from three independent growth curves.

Results

Production of glutathione in CDM

To introduce glutathione producing capability in L. lactis, the NICE system was used. As glutathione is not present in most Gram-positive bacteria (Fahey et al. 1978), and the combined gshA and gshB genes could not be found in the currently available genome sequences of Gram-positive bacteria, gshA and gshB genes were amplified from E. coli, cloned and introduced into L. lactis NZ9000. Figure 1 shows that, after nisin induction, a high amount of γ-GC was produced in strain NZ9000 (pNZ3201), while a small amount of glutathione was produced in strain NZ9000 (pNZ3203). There was no detectable glutathione in cell extracts of strain NZ9000 (pNZ3201) and strain NZ9000 (pNZ3202), indicating that L. lactis NZ9000 lacks γ-GCS and GS activities.

Fig. 1
figure 1

Intracellular thiol content of Lactococcus lactis NZ9000 harbouring different plasmids grown in chemically defined medium (CDM) (not shown) or M17 broth (illustrated here). Cells were harvested after 4 h induction with 2 ng ml−1 nisin. Cys Cysteine, gamma-GC γ-glutamylcysteine, GSH glutathione

Full size image

To evaluate the expression level of γ-GCS and GS, L. lactis NZ9000 harbouring the empty vector pNZ8148 (as a control), pNZ3201, pNZ3202 and pNZ3203 were grown under nisin inducing conditions, and extracts of the cultures were prepared and analysed by SDS-PAGE (Fig. 2). Growth of strain NZ9000 (pNZ3201) and NZ9000 (pNZ3203) in the presence of nisin resulted in the appearance of an additional protein band with an apparent molecular mass of approximately 58 kDa, which is the expected size of γ-GCS. In addition, growth of strain NZ9000 (pNZ3202) and NZ9000 (pNZ3203) in the presence of nisin resulted in the appearance of an additional protein band with an apparent molecular mass of approximately 36 kDa, which is the expected size of GS. The high and functional expression of gshA in strain NZ9000 (pNZ3201) resulted in a high level of γ-GC production (Fig. 1), and the expression of gshA and gshB in strain NZ9000 (pNZ3203) resulted in the production of GSH. Remarkably, the intracellular level of γ-GC in strain NZ9000 (pNZ3203) was only half that of strain NZ9000 (pNZ3201), and the level of GSH in strain NZ9000 (pNZ3203) was very low (Fig. 1). This might be due to the fact that the expression levels of gshA and gshB in strain NZ9000 (pNZ3203) were lower than those in strains with the individually expressed genes (Fig. 2). In addition, the expression level of gshB in strain NZ9000 (pNZ3202) was significantly lower than that of gshA in strain NZ9000 (pNZ3201), suggesting that the presence of gshA upstream of the gshB gene decreased the efficiency of transcription.

Fig. 2
figure 2

Coomassie blue-stained gel after sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE) of cell-free extract (CFE) of L. lactis NZ9000 grown in CDM in the presence (lanes 2–5) or absence (lane 6) of nisin. Lanes: 1 Protein standards (molecular masses in kDa indicated on the left), 2–6L. lactis NZ9000 harbouring pNZ8148 (2), pNZ3201 (3), pNZ3202 (4), pNZ3203 (5, 6). Additional bands resulting from nisin induction and representing γ-glutamylcysteine synthetase (γ-GCS) protein (lanes 3, 5) and glutathione synthetase (GS) protein (lanes 4, 5) are indicated by arrows

Full size image

Increased production of glutathione by L. lactis ssp. cremoris NZ9000 (pNZ3203)

The low intracellular level of glutathione in strain NZ9000 (pNZ3203) grown in CDM might be due to the poor nutrient supply in CDM. Therefore, instead of CDM, M17 broth was used to assess the overproduction level of GSH in strain NZ9000 (pNZ3203), since strain NZ9000 does not take up GSH present in M17 broth. L. lactis NZ9000 is a derivative of L. lactis ssp. cremoris MG1363, which was demonstrated to lack the capability of taking up GSH from the environment in a previous study (Li et al. 2003). Growth in M17 broth with nisin induction resulted in a nearly 8-fold increased GSH level compared to that of the culture grown in CDM (Fig. 1), suggesting that nutrient composition can greatly affect the production of glutathione.

Remarkably, the intracellular molar ratio of GSH to γ-GC of the culture grown in CDM (1:9) decreased to nearly 1:1 when cells were grown in M17 broth (Fig. 1). The latter ratio was still unusually low, since γ-GC, as an intermediate, does not accumulate to such a high level in E. coli (Carmel-Harel and Storz 2000). As a first approach to increase the production of GSH on CDM, the three precursor amino acids of glutathione, namely l-glutamate (Glu), l-cysteine (Cys), and glycine (Gly) were added to the medium. It has been reported that addition of precursor amino acids enhanced the production of GSH by E. coli (Li et al. 1998). In L. lactis NZ9000 (pNZ3203), upon addition of 5 mM and 10 mM of the three precursor amino acids, the levels of GSH appeared to be increased 6-fold and 8-fold, respectively, relative to the level without added amino acids (Table 2). More importantly, the ratio of GSH to γ-GC increased along with the increase of the concentration of mixed amino acids. A very high ratio of 120:1 was achieved upon addition of 10 mM mixed amino acids (Table 2), suggesting that addition of precursor amino acids significantly stimulated the conversion of γ-GC to GSH.

Table 2 Effect of addition of amino acids mixture on production of thiol by L. lactis NZ9000 (pNZ3203). Cells were grown in M17 broth and harvested after 6 h induction with 2 ng ml−1 nisin. γ-GC γ-glutamylcysteine, GSH glutathione
Full size table

To study which amino acid was responsible for the increased production of GSH, different combinations of Glu, Cys, and Gly (5 mM of each) were added to nisin-induced cultures. Figure 3 clearly demonstrates that the increased production of GSH upon addition of mixed amino acids can be fully attributed to the addition of cysteine. Although glycine might be expected to be a putative stimulator of the conversion of γ-GC to GSH, this was shown not to be the case (Fig. 3). The stimulating effect of cysteine on the conversion of γ-GC to GSH is remarkable, since cysteine itself is not involved in the conversion process of γ-GC to GSH. However, analysis of γ-GCS and GS activities in strain NZ9000 (pNZ3203) revealed the interesting phenomenon that GS activity increased by roughly 4-fold and 5-fold when 5 mM and 10 mM cysteine, respectively, were added to the culture, while the activity of γ-GCS remained at a high level (Table 3). The increased activity of GS upon cysteine addition may account for the increased conversion of γ-GC to GSH, but the mechanism behind this effect remains unknown.

Fig. 3
figure 3

Effect of different combinations of glutamate, cysteine, and glycine (5 mM of each) on the production of thiol by L. lactis NZ9000 (pNZ3203)

Full size image
Table 3 Activities of γ-GCS and GS in L. lactis ssp. cremoris NZ9000 (pNZ3203) grown in M17 broth supplemented with different concentrations of cysteine. Cells were grown in M17 broth and harvested after 6 h induction with 2 ng ml−1 nisin.
Full size table

Expression of γ-GCS enhances cysteine uptake

The expression of γ-GCS in strain NZ9000 (pNZ3201) and the expression of γ-GCS and GS in strain NZ9000 (pNZ3203) resulted in intracellular levels of cysteine increased 20-fold and 10-fold, respectively, compared to the control (Fig. 1). In addition, the effect of adding mixed amino acids on the production of GSH by strain NZ9000 (pNZ3203) also indicated that the intracellular level of cysteine increased (up to 13-fold) upon addition of 10 mM mixed amino acids (Table 2). Therefore, it would be interesting to know whether the increased intracellular level of cysteine was caused by increased biosynthesis or increased uptake capability. To elucidate this phenomenon, strain NZ9000 (pNZ3203) and strain NZ9000 (pNZ8148) were grown in CDM free of nisin and no growth difference was found (Fig. 4). When nisin was added to the medium at the beginning of cultivation, the growth of strain NZ9000 (pNZ8148) was almost unaltered, while the growth rate of strain NZ9000 (pNZ3203) was severely decreased (Fig. 4), possibly due to the protein burden. However, the addition of 0.5 mM and 5 mM cysteine partly restored the growth of strain NZ9000 (pNZ3203) in the presence of nisin, indicating that the cells took up more cysteine from the medium to satisfy physiological demand. The effect also demonstrates that the capacity for cysteine biosynthesis limits growth in strain NZ9000 (pNZ3203). Furthermore, when grown in medium supplemented with 5 mM additional cysteine under nisin induction, the intracellular cysteine level of strain NZ9000 (pNZ3203) (±100 nmol mg−1 protein) was 2.5-fold higher than that of strain NZ9000 (pNZ8148) (±40 nmol mg−1 protein). This result indicated that the expression of gshA and gshB significantly enhanced cysteine uptake.

Fig. 4
figure 4

Growth of L. lactis ssp. cremoris NZ9000 (pNZ8148) (circles) and NZ9000 (pNZ3203) (squares) in CDM in the absence (closed symbols) and presence (open symbols) of 2 ng ml−1 nisin. Triangles and rhombuses represent the growth of strain NZ9000 (pNZ3203) in the presence of 0.5 mM and 5 mM cysteine, respectively. Nisin was added to the medium at the beginning of cultivation and growth was monitored in a kinetic microtitre plate reader. The pre-culture was cultivated in CDM without nisin

Full size image

Glutathione does not protect strain NZ9000 (pNZ3203) against oxidative stress

To investigate whether the GSH produced in strain NZ9000 (pNZ3203) leads to an increased resistance to oxidative stress as shown in L. lactis ssp. cremoris SK11 (Li et al. 2003), cells of NZ9000 harbouring plasmid pNZ8148, pNZ3201, pNZ3202, and pNZ3203 were grown in CDM and induced by 2 ng ml−1 nisin. Cells induced for 4 h and 20 h were washed and exposed to 10 mM H2O2 for 15 min followed by measurement of the survival rate of colony forming units. Unexpectedly, strain NZ9000 (pNZ3203) did not exhibit increased resistance to treatment with H2O2 (data not shown). On the contrary, the resistance to H2O2 treatment of strain NZ9000 (pNZ3203) was slightly decreased compared to that of strain NZ9000 (pNZ8148). To investigate whether the resistance of strain NZ9000 (pNZ3203) to H2O2 treatment is growth phase dependent, cells grown in difference phases after nisin induction were withdrawn and exposed to H2O2 treatment, upon which the growth of subcultures was monitored. The duration of the lag phase of strain NZ9000 (pNZ3203) in subcultures was longer than that of strain NZ9000 (pNZ8148), independent of the growth phase (data not shown). This confirms that the GSH produced in strain NZ9000 (pNZ3203) does not lead to a higher level of resistance to oxidative stress.

Discussion

L. lactis has become a routine prokaryotic host for metabolic engineering, and a variety of strategies have been proven to be successful (Kleerebezem and Hugenholtz 2003). The biosynthesis of GSH is a relatively simple pathway but is not present in L. lactis. The genes encoding γ-GCS and GS cannot be found in the currently available genome sequence of L. lactis ssp. lactis IL1403 and experimental evidence for the lack of GSH biosynthetic capacity was provided in a previous study (Li et al. 2003). However, some strains of L. lactis do have the ability to take up glutathione from the environment. Glutathione taken up by L. lactis ssp. cremoris SK11 increases resistance to H2O2 stress (Li et al. 2003).

In this study, GSH biosynthetic capability was introduced into L. lactis ssp. cremoris NZ9000 by over-expressing the gshA and gshB genes from E. coli using the NICE system. Cloning of gshA and gshB genes from E. coli is not new but the attempt to express these two genes in Gram-positive bacteria has never been made. Nearly 20 years ago the gshA and gshB genes were separately cloned from E. coli B (Murata et al. 1983; Murata and Kimura 1982) and sequenced (Gushima et al. 1984; Watanabe et al. 1986a). Although the introduction of the hybrid plasmid pGS500 carrying both gshA and gshB genes into E. coli RC912 resulted in a simultaneous increase in the activities of γ-GCS (10-fold) and GS (14.5-fold), the intracellular GSH concentration in E. coli RC912 (pGS500) cells increased only 1.3-fold compared to the wild type (Gushima et al. 1983b). The reason for this disappointing result was ascribed to the fact that the activity of γ-GCS in E. coli is feedback inhibited by GSH. The GSH concentration required for 50% inhibition of γ-GCS, 2.5 mM (Watanabe et al. 1986b), is comparable to the intracellular level measured in stationary phase cells of E. coli B (approximately 3 μmol ml−1 packed cells) (Gushima et al. 1983b; Murata et al. 1981). A similar phenomenon was observed in recombinant Saccharomyces cerevisiae over-expressing gshA and gshB genes from E. coli B. There, the intracellular glutathione content increased only 2-fold although the expression of γ-GCS and GS increased 1,039-fold and 33-fold, respectively (Ohtake et al. 1988, 1989).

Upon addition of 10 mM cysteine, the activities of γ-GCS and GS in L. lactis NZ9000 (pNZ3203) were 10-fold and 4-fold, respectively, of those measured in E. coli TG1 (Table 3), indicating that both genes were over-expressed in L. lactis. However, compared with the enzyme activities in E. coli RC912 (pGS500) cells calculated from literature (Gushima et al. 1983b), the activities of γ-GCS and GS in L. lactis NZ9000 (pNZ3203) were not exceptionally high: the activity of γ-GCS was 2.7-fold that in E. coli RC912 (pGS500) while the activity of GS was only one-fifth that in E. coli. In spite of this, the intracellular GSH concentration in L. lactis NZ9000 (pNZ3203), 358 nmol mg−1 protein (approximately 140 mM, assuming an intracellular volume of 2–3 μl per milligram of cell protein, Table 2), was roughly 20-fold of that of E. coli RC912 (pGS500) (Gushima et al. 1983b) and 3-fold of that of S. cerevisiae YNN27 (pGSX120E) (Ohtake et al. 1989). This unexpectedly high concentration of GSH in L. lactis NZ9000 (pNZ3203) indicates (1) that there is no γ-glutamyltranspeptidase activity present in L. lactis (in E. coli and S. cerevisiae this enzyme degrades GSH); (2) the feedback inhibition of GSH on γ-GCS seems not to be effective in L. lactis NZ9000 (pNZ3203), since the GSH concentration of 140 mM is obviously higher than the concentration required for feedback inhibition. The latter phenomenon is interesting, since it has been proposed that feedback inhibition of γ-GCS by GSH is the rate-limiting step in glutathione biosynthesis (Anderson 1998; Huang et al. 1988). There are three possible interpretations: (1) the glutathione produced by L. lactis is an oxidised form (GSSG) or bound to protein. This possibility was excluded because the presence of dithiothreitol (a strong thiol reducing agent) in the pre-treatment buffer for thiol analysis did not affect the GSH concentration measured by HPLC (data not shown); (2) the structure of γ-GCS was modified by the host, which is an unlikely possibility; (3) the unexpectedly high intracellular concentration of cysteine, caused by increased cysteine uptake, releases (or desensitises) γ-GCS from feedback inhibition by GSH.

Cysteine availability is normally the key rate-limiting factor in glutathione biosynthesis, and the maintenance of adequate intracellular GSH levels is dependent upon the extracellular availability and transport of cysteine into cells (Anderson 1998). In this study, we found that expression of gshA enhanced the activity of the cysteine transport system of L. lactis, although the exact mechanism is unknown. More strikingly, the activity of GS in L. lactis was improved by increasing the extracellular cysteine availability. SDS-PAGE did not show significant differences in GS levels under different cysteine concentrations (data not shown).

The GSH produced by L. lactis NZ9000 (pNZ3203) does not increase resistance to H2O2 treatment. Since strain NZ9000 neither produces nor imports glutathione, it possibly lacks the enzymatic machinery to use glutathione for the reduction of oxidative stress. This might explain why NZ9000 does not benefit from glutathione production. Alternatively, a strain like SK11 that is known to benefit from glutathione could be used as a host to produce glutathione.

The strain or its method of construction described here can be used to investigate the effects of in situ glutathione production by lactic acid bacteria on properties of fermented foods like cheeses, where beneficial effects of the addition of glutathione have already been shown (Singh and Kristoffersen 1971).